The Glitch, The Glitch, Oh The Itch; The Media At a Fever Pitch

Each time there is a technology glitch in our modern markets, media reaches out to us, and we kind of tell them the same thing… it is a by-product of complexity, and glitches will continue to happen.

Yesterday, as you know well, NYSE experienced system issues that started very early in the day, ultimately resulting in a NYSE-only trading halt that lasted hours. While some have drawn comparisons with the August 2013 NASDAQ SIP outage, that outage involved a regulatory halt that shut trading down across all systems. By the way, this NYSE system outage follows on the heels of a BATS (EDGX) system outage that lasted 30 minutes on Monday July 6th.

NYSE’s first alert to the street went out just after 8:00 am; there was an issue with a gateway connection. Despite that gateway connection issue, NYSE opened normally, yet began to notice that customers “were not receiving some messages.” NYSE made the decision to halt trading on NYSE at 11:32 am. They found the issue, corrected it, and ultimately re-opened by 3:13pm – in time for the closing auction.

NYSE’s system issue is suspected to be related to the first leg of a NYSE system upgrade that began implementation this week.

While we all often talk about market fragmentation as a bad thing (actually we all do – including other buyside, brokerage, and market making firms), today it worked out well for long term investors. Trading in NYSE stocks shifted over to the other public stock exchanges, as well as ATSs. The shift was seamless, and many of us at Themis, as well as our clients, remarked how smooth and uneventful the “NYSE glitch” was that was dominating the media airwaves.  While it may have been a non-issue for most low-frequency traders, as we don’t tend to send in millions of orders, cancellations, and revisions in sub-second times periods, it certainly was an issue for higher frequency players. Likely, if you were a high frequency player missing messages, and unsure of your inventory and positions, the NYSE glitch was a very big deal.

Also, this outage would have been a very big deal for everyone if NYSE did not re-open for the closing auction, as the NYSE closing price is relied upon to price thousands of stocks, ETFs, mutual funds, and portfolio statements. While in the past, synthetic proxies have been created when a primary exchange failed to run its closing auction, the closing prices attached to the real closing auction are superior, and generally more trusted.

Additionally, this NYSE outage may have affected trading more than the general media understands; let me explain what I mean:

1)      Trading today is complex – involving 50-plus destinations, and sophisticated order-routing technology. In other words, the buyside often uses algorithmic technology to break up and piece out large orders systematically (VWAP, TWAP, POV, etc.)


2)      Those algorithmic orders had the real potential to cause more havoc, had NYSE not reopened and ran their closing auction. A 500,000 share buy order, that would have allocated some large percentage to “the close” for execution, would likely have shifted that trading volume to the 15-20 minutes or so prior to 4:00pm. This would likely make little difference in a stock like BAC, but a potentially greater dislocative difference in a more thinly traded name.


Some experts have been very vocal in their criticism of NYSE:

–          “How can NYSE allow a system upgrade without sufficient testing?”

–          “Why didn’t NYSE fail over to its Disaster Recovery (DR) protocols and location?”

–          “Didn’t’ the SEC pass a rule that prevents this sort of outage and issue?”

Actually, the SEC did pass a set of rules – titled Reg SCI – in February of this year. Compliance for that set of rules begins in November of this year, so those rules do not apply – yet. Come November though… we might all see a very intolerant regulator.

We would like to leave you with two thoughts on NYSE’s system issue yesterday.

First, there is a large transparency and regulatory disparity between exchanges and dark pools. Dark pools have system issues at least as frequently as exchanges; you just don’t hear about them… because they’re dark! We would venture to guess – and this is a generalization – that stock exchange technology is generally more robust, faster, and better-maintained than dark pool technology. So, as we demand improvements from NYSE and other exchange operators, so should we from our dark pools and ATSs.

Second, the SEC’s Reg SCI is 743 pages long. In comparison, Reg NMS is just a paltry 523 pages long! While perhaps well-meaning, and designed to give investors a degree of confidence, so that they may trust our very complex trading environment nowadays, rules and legislation will never be able to stop technology from failing.

We will have glitches and outages in the future. For-profit trading systems will invest in maintenance and performance to the extent that their profits allow. Some venues will be better at it than others. This is inescapable.

So, be prepared. Next time, it might come on a quad witching. Or an S&P rebalance day. Or a Fed day. And while most of you can easily afford to sit back and watch, and not play during a period of mayhem, some industry players will not know their trading positions. And that uncertainty will exacerbate the outcome.